Step-by-step instructions on how to analyze your social media content (and data) to find out what types of content are most effective, which posts and words drive specific actions and how to use your data to answer any other questions you have about your content (best time to post, best post length, etc.). This method can easily be applied to any social media content (Facebook, Twitter, Instagram, Pinterest, etc.) to help you become a more effective, efficient digital marketer.
2. Pop Quiz
Do you know the answers to these questions about your brand’s content across social
channels?
What call-to-action keywords do your Facebook fans respond to most?
Are fans more likely to retweet content about a buy one get one free promo or a 50% off
sale?
To maximize engagement, should your Instagram photos be extreme close-ups or should
you use a tool like PicStitch to splice different photos together?
3. Answers
Industry reports are great, but aren’t specific to your brand
I’ve seen some clients with data that is nearly an exact opposite of suggested “best practices”
Insights can make you more effective and efficient as a content marketer
You can calculate them yourself with MS Excel, Pivot Tables and a little elbow grease. I’ll
show you how
5. Collecting and Organizing the Data
Download your page’s post‐level data from Facebook Insights for the past six months
I recommend going back two weeks, then pulling the previous six months
Create a new spreadsheet and splice together the data that matters most to you for your
goals. Some basics to begin with:
Reach
Impressions
Engagement Rate (Engaged Users/Reach)
Negative Feedback Rate (Lifetime Negative Feedback/Reach)
6. Editing Columns
Delete columns with information you’re not interested in, add columns for Engagement
Rate and Negative Feedback Rate, change column headers and delete descriptions –
whatever makes the most sense to you
7. Formatting Dates and Times
Add four additional columns to the right of the Posted column. Title them Day of week,
Date, Time and Rounded
Use the following formulas to calculate the first value in each column (replace CELL with the
correct cell info)
Day of week: =TEXT(CELL,”dddd”)
Date: =INT(CELL)
Time: =MOD(CELL,1)
Rounded: = ROUND(CELL/(1/24),0)*(1/24)
You’ll probably have to format each of the columns to get them to display correctly. Just select
the column, right click and choose Format Cells
If it all looks good, drag the formula down to fill the rest of the cells
8. Formatting Dates and Times
The four new columns should look something like this:
9. Calculating Post Length
Add a column to the right of the Post Message column and title it Characters
In the first row, type =LEN(CELL) where CELL is the appropriate cell
If that gives a character count, drag the formula down to the rest of the cells
Add another column to the right of the Characters column and title it Post Length
Use a LOOKUP table to define ranges. If you’re not familiar with the LOOKUP function, here’s an
explanation. You can also use the example on the next slide and substitute your own numbers
10. Calculating Post Length
Use a LOOKUP table to define
ranges for post length. Going
smaller than intervals of 10
probably won’t give any
actionable insights, so don’t
go overboard
11. Adding Labels
Add a few columns for things you want to track (campaign name, purpose of post,
topic, call to action style, etc.)
Think of these columns as buckets. You need to have enough content to fill all of the
buckets, so try not to get too granular
12. Bonus Points: Adding Web Analytics
If you’re using UTM tags in the links you share on social sites, hopefully they’re pulling in
conversion and/or goal data for each post (whether you’re an e-commerce site or not). If
you have it, take the time to add it into your spreadsheet. Tedious, but extremely helpful
(you’ll see an example later in the deck). I recommend adding:
Unique Visits
Conversion Rate or Goal Completion Rate
Not sure why or how to track conversions or goals? Here’s a great article to get you started.
13. Building Pivot Tables
Now, you can create Pivot Tables to help you “bubble up” what’s working and what isn’t.
From your spreadsheet:
Click Insert>Pivot Table
Highlight the source data (should be everything on your spreadsheet, except for the
LOOKUP table) and choose to have the table placed in a new spreadsheet
Drag and drop items from the fields list to areas in the table builder (screenshot on next
slide)
Play around with different ways of mashing up the data. It’s always a good idea to add a
“Count of…” column that counts some unique identifier for each post (in this example, it’s
Post ID), so you can make sure you’re not working with really small or skewed fields
14.
15. Building Pivot Tables
• If you’re not sure what the data “looks” like, you can toss it into a PivotChart and see if
that helps
17. Discover Effective CTA Styles and Words
If the goal is engagement, explicit calls to action are more effective than implicit CTA’s
Using the phrase “let us know in the comments” generates more engagement than “like if.”
Implicit CTA’s with the words “when” and “how” typically drive more engagement
Using “like if” and “share if” result in 7x more negative reactions from fans (hiding content, unliking the page, etc.)
Applying the insight: Use more explicit calls to action, but use “like/share if” CTA’s with caution. Ask more questions using the words
“when” and “how”
18. Test The Rule of Thirds
We can see that curated content is posted less frequently, but has a higher engagement rate
When the brand posts content about the fans, they comment and convert at a higher rate. Specifically, posts about
guest bloggers get higher click rates and conversion rates
Posts about home‐related topics and travel are highly shareable and engaging content
Posts about toys generate an abnormally high amount of negative feedback
Applying the insight: Post more curated content to increase engagement rate (and EdgeRank). Working in more guest blogger
content will help drive sales and creating more posts about home and travel ‐related topics can help drive engagement. Consider
reducing or eliminating posts about toys to reduce negative feedback (which hurts EdgeRank)
19. Find Topics That Resonate With Your Fans
Posts about the animal shelter charity effort got far more engagement than the posts about the Earth Day efforts
People are more likely to click on posts about productivity, but are less likely to share those posts
Applying the insight: In the future, look to do more partnerships with animal ‐focused charities. You may want to try posting
non‐charity content with animals in it to see if just adding animals increases engagement. Continue posting on productivity and
organization, but consider doing a survey of fans/readers to see why they aren’t sharing the productivity articles. Is it because they
don’t find them useful? Or possibly because they don’t want to share productivity “secrets” with others (co ‐workers, competitors,
etc.)?
20. Choose the Best Content for Each Goal
Let’s assume that the purpose of our presence on Facebook is to do four things: build brand awareness, engage with
fans, drive online sales and establish ourselves as a thought leader in the industry
For sales, a link is the best converting content. To help us build our brand, we want as many eyes as possible on our
content and photos get the most impressions. The best content types to engage our fans are photos and video
Applying the insight: Use the best content type for each goal more frequently, but continue to post alternate content types to see
if there is a shift in the most effective content type (Ex. 50% of sales posts should be a link and the other 50% should be a mix of
photo and video)
21. Application on Additional
Platforms
Utilizing the labeling strategy across social and web properties
22. Twitter
Do links get more RT’s than comments?
What’s the optimal tweet length?
Do tweets with hashtags get more engagement?
Take it Anywhere Do links with vague descriptions get more clicks?
Instagram
You can take this labeling strategy
and apply it across platforms. Get Do close‐up shots do better than PicStitch shots?
Twitter data (from tools like Sprout Do posts with hashtags get more engagement?
Social, Hootsuite), Instagram stats Do photos of people do better than product shots?
(from tools like Statigram, Ink360),
Pinterest data (from tools like Pinerly, Do photos of text get more engagement than regular photos?
Curalate) and figure out which
questions are most important to you,
Pinterest
then add labels to your content to
help you answer the questions. Do photos of clothes on models get more engagement than
product shots?
Do photos from Instagram do as well as photos from our website?
Here are a few examples of questions Do photos with price info convert at a higher or lower rate than
you could answer with content labels: photos without price info?
Do small infographics get repined as frequently as large ones?
24. Potential Problems
Beware of small data sets. If you’re trying to determine which content type is most
effective for curated content and you only have a few posts of each type, it’s really easy
to get skewed data. Below is an example of how one post can throw an entire table off.
Ideally, you’d have more posts of each type so you’re not relying on such a small data set. You
could also use the median instead of the average to prevent skewed data in your pivot table.
Unfortunately, Excel doesn’t offer Median as one of the standard calculations in Pivot Tables, so
you’ll need to hack one together using arrays.
25. Check Yourself
Define your questions, then review the numbers. If you’re even remotely mathematically
inclined (or just really love Excel), you can absolutely cripple yourself with the possibilities and
different ways of looking at things, so be very specific in what you’re looking for
Data is important, but so are your instincts. If you know that posting about current political
events always generates high engagement, but creates a hostile environment on your page,
go with your gut and scale back on the political content
Correlation does not imply causation. Looking at data and crunching numbers can be pretty
cool, but as with any “experiment,” it’s important to remember that LOTS of factors can
influence if and how people interact with your content. Always look for underlying stories or
factors that may have influenced the numbers
I like to review the insights I find and pretend like I’m someone from a competing company whose single
goal is to poke holes in my theory – ask (and answer) the hard questions. You’ll be better for it
Test, refine, repeat. What worked a few months ago might not still work today. Make sure you’re
looking at your data by date (in addition to label) and always testing new theories
A good ratio to shoot for is 80% content you know “works” and 20% “test” content to help you find new
insights
26. Author:
Go forth and create
Courtney Livingston
good content. Digital Marketing Strategist
courtneylivingston.com
And if you have any questions or
suggestions, please reach out!